Role of Ryanodine Receptors in the Assembly of Calcium Release Units in Skeletal Muscle

The methods used to create the 1B5 cell line are described in detail elsewhere (Moore et al., 1998) but, briefly, it was done in four stages:

CCS ES cells that had successfully undergone homologous recombination disrupting the ry₁r gene at nucleotide (nt) 840 in exon 10 underwent a second round of selection using high G418 (3.2 μg/ml) to select for the rare homologous recombination of the disrupted allele giving ES cells where both alleles are disrupted (Mortensen et al., 1992).

ES cells that were homozygous for the disrupted ry₁r allele were injected subcutaneously into hind quarters of SCID mice at a concentration of 2–5 × 10⁶ ES cells/cm³ in a volume of 1 cm³. The injections resulted in ry₁r−/ry₁r− teratocarcinomas composed of several cell types, including 5–10% skeletal muscle. The teratomas were minced, digested with trypsin, and then the cells were plated in growth medium (DME; with 20% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and additional 2 mM l-glutamine).

Several primary ry₁r−/ ry₁r− fibroblast clonal lines were established.

Several of the ry₁r−/ ry₁r− fibroblast lines were infected with a retrovirus containing a puromycin myoD construct. 48 h after infection, selection was started with puromycin (1 and 2 μg/ml) and clones were picked after 7–10 d. Puromycin- resistant clones were subcloned in growth media. One of the subclones, the 1B5 cell line, was chosen for complete characterization based on its characteristics. It exhibited contact growth inhibition at high density in growth medium and could be induced to differentiate into noncontractile multinucleated myotubes when 5% heat-inactivated horse serum (HIHS) was substituted for 20% FBS in DME medium.

The cells were grown at 37°C in growth medium. After 2 d, the cells were replated on thermanox coverslips (Nunc Inc., Naperville, IL) covered with Matrigel (Collaborative Biomedical Products, Bedford, MA). The medium was changed every 2 d. At ∼70% confluence, the medium was replaced by a low serum medium (no FBS) containing 5% HIHS to induce differentiation. The differentiation medium was changed every day and the cells were fixed 4–6 d later.

The cells were transfected during the growing phase on thermanox coverslips when they reached ∼80% confluence. A transfection mixture containing 20 μl of lipofectamine and 3.6 μg of cDNA (in 0.1 ml of Optimem medium) per dish to be transfected was prepared 1 h before transfection and kept at room temperature. Immediately before transfection, 1 ml of Optimem medium was added to the mixture. The growth medium in the dishes was replaced with the transfection mixture and an equal amount of differentiation medium was added 6–8 h later. The transfection mixture was removed 15–20 h later and replaced with differentiation medium. The medium was changed every day and the cells were fixed 4–5 d later.

Cultures grown on coverslips were fixed either in methanol or acetone for 20 and 10 min, respectively at 20°C, blocked in PBS containing 1% BSA and 10% goat serum for 1 h, and then incubated in primary antibodies overnight at 4°C. After washing four times for 10 min in PBS/BSA, the coverslips were incubated in either FITC-conjugated (Cappel Laboratories, Malvern, PA) or CY3-conjugated (The Jackson Laboratories, Lexington, KY) goat anti–mouse secondary antibodies for 1 h at room temperature, and then washed again. The coverslips were mounted in glycerol containing 0.0025% para-phenylenediamine, 0.25% 1,4 diazobicyclo 2,2,2 octane, and 5% N-propylgallate to retard photobleaching. The working dilutions and the sources of primary antibodies are listed in Table I. All antibodies used have been previously characterized. Control slides were not incubated in the primary antibody. The specimens were viewed on a scanning confocal microscope (MRC-600; Biorad Laboratories, Hemel Hempstead, Hertfordshire, England).

The cells were washed twice in PBS at 37°C, fixed in glutaraldehyde, and then kept in fixative for up to 1–4 wk before further use. For thin sectioning, the cells were fixed in 3.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.2, followed by 2% OsO₄ for 2 h at room temperature, and then saturated uranyl acetate for 4 h at 60°C. The samples were embedded in Epon 812, and the sections stained in saturated aqueous uranyl acetate followed by lead salts, both for 8 min.

For freeze fractures, the cells were fixed in glutaraldehyde as above, and infiltrated in 30% glycerol. A small piece of the coverslip was mounted with the cells facing a droplet of 30% glycerol, 20% polyvinyl alcohol on a gold holder, and then frozen in liquid nitrogen–cooled propane (Cohen and Pumplin, 1979; Osame et al., 1981). The coverslip was flipped off to produce a fracture that followed the culture surface originally facing the coverslip. The fractured surfaces were shadowed with platinum either at 45° unidirectionally, or at 25° while rotating, and then replicated with carbon in a freeze fracture (model BFA 400; Balzers S.p.A., Milan, Italy). Sections and replicas were photographed in a EM 410 (Philips Electron Optics, Mahwah, NJ).

The density of foci in confocal images was measured using the following procedure: (a) an area covering 100 μm² was selected based on the sharpness of the immunolabeled spots. This selection procedure accepts areas in which the optical section grazes the surface membrane and excludes areas where the focal plane is below the cell surface and the spots are diffuse; (b) the contrast was increased to bring the spots to saturation. Under these conditions the large majority of spots are very well defined (see detail in Fig. 7,B); (c) the spots were counted; and (d) two or three measurements were performed in different areas for each cell. For more details about foci counting see Fig. 7 legend.

The width of the junctional gap between SR and surface membranes was measured along lines drawn at several random positions across the junctions, using a dissecting microscope fitted with an eyepiece micrometer.

The density of large membrane particles was calculated from counts of all junctional domains and several randomly selected nonjunctional domains in micrographs with excellent shadow quality (comparable to the one shown in Fig. 6).

The apparent diameter of intramembranous particles was measured at right angle to the direction of platinum deposition. The thickness of platinum deposition is closely monitored using a quartz thin film monitor; thus variations in apparent particle diameter because of this factor are small. The approximate particle height was determined by measuring the length of the platinum-free “shadow.” The length of the shadow is influenced by the angle of platinum shadowing and this in turn is affected by curvature of the shadowed membrane. Variations resulting because of shadowing angle are minimized by avoiding micrographs with obviously foreshortened or elongated shadows. Particles clustered within junctional domains and visually identified as large were measured until a count of 100 was reached. All particles within regions of approximately equal size, but showing no sign of large-particle clustering, were measured until the same total count was reached.

The density of large particle clusters on the surface membrane was calculated from counts of electron micrographs at magnifications between 20,000 and 44,500. The density of immunoreactive spots was calculated from counts of digitized images enlarged ∼10,000 fold.

1B5 cells cultured in presence of growth factors are usually small, fusiform, and mononucleated (Fig. 1,A). After withdrawal of growth factors, most cells fuse and differentiate into aligned multinucleated myotubes, some 50 μm or more in width (Fig. 1 B).

Differentiated 1B5 cells were immunolabeled with antibodies specific for junctional proteins of exterior membranes (α₁ and α₂ subunit of the DHPR) and SR (triadin). 5 d after withdrawal of growth medium, about a third of the myotubes are positive for these antibodies. α₁- and α₂-DHPR and triadin are clustered in discrete foci, similar to those found in normal skeletal developing muscle and in adult skeletal and cardiac muscle (Figs. 2 and 3; compare with Yuan et al., 1991; Flucher et al., 1993b, 1994; Knudson et al., 1993a; Guo et al., 1994, 1996; Carl et al., 1995a,b). The majority of immunopositive spots are located on or near the surface membrane, on both ventral and dorsal sides of the myotubes (Fig. 2; see Fig. 3 for a through focus series). The images cannot distinguish between foci of immunofluorescence at the surface of the cell and in the cortical cytoplasm (Yuan et al., 1991). The ventral side of the cell, facing the coverslip, is flat and thus gives the best images (Fig. 2).

Since the two antibodies used for the immunochemistry are both derived from mouse, we did not attempt double immunolabeling. However, several observations indicate that the location of DHPR- and triadin-positive foci coincide. First, the density of triadin-positive foci at or close to the coverslip-facing side of the cell (35.8 ± 9.6/100 μm² from 65 measurements in 25 cells, 23 confocal images, mean ± 1 SD) is very similar to the density of DHPR foci (38.8 ± 7.0/100 μm² from 102 measurements in 33 cells, 34 confocal images). Second, when the cells were colabeled with both antibodies (anti–α₁-DHPR and anti-triadin) the density of foci at or close to the coverslip-facing side of the cells is practically identical to that obtained by single labeling with either antibody (36.3 ± 5.5/100 μm² from 52 measurements in 17 cells, 16 confocal images). If the two molecules were not colocalized we would definitely expect a higher number of positive foci (approximately double). In addition, the intensity of foci was definitely higher in the case of cells stained with both antibodies than in cells stained in parallel with either antibody alone. A third observation indicating coincidence of DHPR and triadin foci is that they are similarly disposed over the cell surface (see Figs. 7 and 8; and see below).

Differentiating 1B5 cells develop somewhat disordered and defective myofibrils containing variable amounts of myofibril components. More differentiated cells have A bands that tend to be randomly oriented and spaced, and with aligned thick filaments and a clear M line (Fig. 4,A, arrows). There is no evidence of Z lines, and a smaller than normal complement of thin filaments is present (Fig. 4 B).

Primitive transverse (T) tubules (or plasma membrane invaginations) and SR systems are present in all differentiating cells containing developing myofibrils, and are absent in cells that do not show any evidence of myofibrillogenesis. T tubule–like invaginations are located at the fiber periphery, mostly within a subcortical strip of cytoskeleton devoid of myofibrils. These invaginations appear most frequently as apparently empty, wide vesicles with a well- demarcated membrane (Fig. 5, C–E), and less frequently as labyrinthine networks resembling multiple caveolae (Fig. 4 C, arrows) (Ishikawa, 1968; Franzini-Armstrong, 1991). The continuity between these vesicles and the fiber surface can be directly traced in some instances (data not shown). There is no evidence of a mature T tubule network penetrating between the myofibrils and associating with them. Few primitive T tubules (identifiable by their proximity to an SR element, see below) are present in the fiber interior in proximity of the Golgi regions.

Numerous small vesicles with an electron dense content are closely apposed either to the plasma membrane (Fig. 5, A–C), or to short T tubules (Fig. 5, C–E). The junctions thus formed closely resemble peripheral couplings and dyads (Fig. 5,F) of normal developing myotubes. We identify the vesicles as SR elements and their electron-dense content presumably represents calsequestrin (Meissner, 1975). Nonjunctional or free SR is scarce. All visible T tubule profiles have at least one SR vesicle associated with them, and thus the densities of junctions along the plasmalemma and in the subcortical and deeper regions of the cell depend on the extent of T tubule development. The free surface of the cells has numerous, short, developing T tubules. On the free side of the cell, 39% of the junctions are peripheral couplings, and 61% are peripherally located dyads (from a total of 193 junctions, from 26 cells). On the other hand, on the substrate side, T tubules are less numerous and peripheral couplings account for 64% of the junctions, whereas peripherally located dyads are only 36% of the total (from 152 junctions, 25 cells). Junctions within the cell interior are rare (43, or 11%, out of a total of 388 junctions in all locations), and they are usually in small clusters within some, but not all, cell profiles. The location of the junctional SR vesicles at or close to the fiber periphery agrees well with the location of the DHPRs and triadin foci seen by immunohistochemistry, which are at or near the surface of the cell (Fig. 3).

All junctions in 1B5 cells are dyspedic, i.e., “feet” are absent from the junctional gap between SR and exterior membranes, and the gap width is narrow (Fig. 5, A–E). On the average, the junctional gap is 6.5 ± 2.6 nm (mean ± 1 SD, n = 52 junctions, 37 peripheral couplings and 15 dyads, 255 measurements) and thus too small to accommodate cytoplasmic domains of RyRs or feet, which are ∼12-nm wide (Radermacher et al., 1994). The irregularly disposed strands that usually occupy the gap in dyspedic myotubes (Fig. 5, A and D, arrowheads) are clearly composed of a protein different from RyR. The gap in feet-containing junctions from normal myotubes is wider (Fig. 5 F, from a normal in vivo myotube). Gap widths of 10–13 nm have been previously measured in junctions from mouse myotubes developing in vivo (Takekura et al., 1995b) and in vitro (Nakai et al., 1997) fixed and stained as in this study. Junctions with a 10–13-nm gap were not detected in 1B5 cells.

In freeze fracture replicas of undifferentiated cultures the cytoplasmic leaflet of the plasmalemma is smooth and it shows intramembrane particles of variable sizes, without any sign of clustering (Fig. 6 A).

The plasma membranes of large myotubes in differentiated cultures have dimples marking the mouths of caveolae and T tubules (Fig. 6,B, arrows) and plasma membrane domains that contain clusters of large, tall particles (Fig. 6,B, semicircles). Small- and medium-sized particles are largely excluded from these clusters (see details in Fig. 6, C–E). The apparent diameter and height of large particles in plasma membrane domains is 9.9 ± 1.7 nm and 10.9 ± 1.5 nm, respectively (mean ± 1 SD; n = 100, 4 cells, 4 freeze fractures). The apparent diameter and height of particles, including large ones, in patches of membrane adjacent to the cluster-containing domains is 6.8 ± 1.5 nm (n = 100) and 6.5 ± 1.8 nm (n = 100), respectively. Differences between diameters and heights in the two groups are extremely significant (Student's t test; P < 0.0001 in both cases). Note that the large-clustered particles are also relatively tall; their height to diameter ratio is 1:10, whereas the ratio in random particles is 0.95. The average density of large particles in junctional plasma membrane domains is 1,077 ± 228/μm² (n = 63; mean ± 1 SD; n = number of domains), whereas the average density of large particles in nonjunctional regions is 104 ± 57/μm² (n = 82). The differences between the two data are large and extremely significant (Student's t test P < 0.0001) confirming the visual observation of clustering.

In D5–6 cultures 35% of the cells (from a total of 334 cells in 7 cultures, 11 replicas) contain plasma membrane domains with clusters of large particles (Fig. 6, B–E). The surface of the rest of the cells resemble that of cells fixed in the presence of growth medium (Fig. 6 A). We could not determine whether the presence of particle clusters is limited to multinucleated cells.

Foci of DHPRs detected by immunolabeling and clusters of large intramembrane particles are present at comparable densities. The two densities, related to the apparent surface area, are 38.8 ± 7.0/100 μm² (from 102 measurements in 33 cells, 34 confocal images, detected by the α₁ antibody) and 26.5 ± 14.0/ 100 μm² (38 measurements in 17 cells, 38 electron micrographs). We expect that the immunolabeled foci at the fiber periphery include both peripheral couplings and peripherally located dyads, whereas clusters of particles detected in freeze-fracture correspond exclusively to peripheral couplings. Thus the density of foci is expected to be larger than the density of clusters. The measured ratio of fluorescent foci to particle cluster densities on the substrate side of the cell (1.46 from the above data) is slightly lower than the ratio of total peripheral junctions (peripheral couplings plus cortical dyads) to peripheral couplings (1.56 from counts in thin sections). The most probable cause for this minor discrepancy is that some of the fluorescent spots are likely to represent more than a single cluster of particles. Indeed, two or three peaks of density are often seen within some of the larger spots. Equal densities of DHPR foci and particle clusters suggests identity of the two structures. This is further supported by the following observations: (a) clusters of particles and foci of DHPR immunolabeling are present on many but not all differentiating cells; (b) clusters of particles and foci of DHPRs are randomly distributed in most cells; (c) in some cells, foci that immunostain positive for anti-DHPR antibodies are strikingly arranged in longitudinal rows (Figs. 7,B, and 8,B). A similar disposition is also observed for plasma membrane domains containing clusters of particles (Fig. 7,D, arrows and semicircles, and 8 C); and (d) some DHPR foci and particle clusters are so closely spaced that they almost fuse with each other (compare Fig. 8 B with C).

The above analysis suggests that the large, clustered particles in the surface of 1B5 cells are DHPRs. Indeed, the large particles in 1B5 clusters are identical in size with the particles representing DHPRs in the tetrads of BC₃H1 cells (Protasi et al., 1997), and are also quite similar to the slightly smaller particles representing DHPRs in cardiac muscle. The apparent diameter and height of the large, clustered particles in 1B5 cells is 9.9 and 10.9 nm, respectively (see above); the large particles forming tetrads in BC₃H1 cells are 9.7 ± 1.2-nm wide and 10.6 ± 1.8-nm tall (from 100 particles, 4 cells, 4 freeze-fractures); the particles in cardiac muscle are 8.5-nm wide and 9.6-nm tall (Sun et al., 1995).

Foci of triadin detected by immunolabeling are also located in prominent longitudinal rows (Figs. 7,A, and 8 A) and have density equal to DHPR foci (see above). Thus, the three detectable components of calcium release units of 1B5 cells (DHPR and triadin foci, detected by immunolabeling; and large particles, detected by freeze-fracture) appear to be colocalized. Since triadin is a junctional SR protein, this colocalization occurs at the junctions between SR and exterior membranes.

Optical diffraction confirms the random disposition of DHPRs in 1B5 cells. Fig. 9 compares images from rotary shadowed replicas of A an array of tetrads in a peripheral coupling from the BC₃H1 cell line, and C, a group of DHPRs from a 1B5 cell. The optical transformation of the tetrad array (Fig. 9,B) shows evidence for two tetragonal arrangements indexing on the spacing between particles within the tetrad (Fig. 9,B, large arrows) and between the centers of tetrads (Fig. 9,B, small arrows; see Protasi et al., 1997). Note that tetrad arrays in BC₃H1 cells have identical parameters to those of other skeletal muscle both in vivo and in vitro (Franzini-Armstrong and Kish, 1995). The optical transformation of the 1B5 cluster (Fig. 9,D) shows only noise derived from the random arrangement of scattering units. This analysis confirms the visual impression that DHPRs in dyspedic junctions of 1B5 cells are randomly disposed and do not form tetradic arrays. The random disposition of DHPRs in peripheral junctional domains resembles that observed in cardiac muscle myocytes (Sun et al., 1995; Protasi et al., 1996; see Fig. 12 for a diagram).

1B5 cells transfected with the cDNA for Ry₁R were immunolabeled with Ry₁R-specific antibody. 4–5 d after withdrawal of growth medium (D4–5) ∼10% myotubes are positive for Ry₁R. These Ry₁R molecules are tightly clustered in discrete foci, similar to those found in normal developing skeletal muscle (Fig. 10; compare with Flucher et al., 1993b; Flucher et al., 1994). Similar to DHPR and triadin foci, Ry₁R immunopositive spots are mostly located on or near the surface membrane, on both ventral and dorsal sides of the myotubes. Nontransfected cells incubated with the same primary and secondary antibodies and transfected cells incubated with only the secondary antibody are negative.

Freeze fracture of transfected cells shows restoration of tetrad arrays in the surface membrane of ∼10% 1B5 cells. Fig. 11, A–C shows clusters of large particles (DHPRs) in the surface membrane of a dyspedic 1B5 cell (Fig. 11,A; see also Fig. 6), of a normal developing myotube (mouse) in vivo (Fig. 11,B), and of a transfected 1B5 cell (Fig. 11,C). In 1B5 cells, the particles have a random arrangement, whereas in normal myotubes and in transfected cells the particles are clearly arranged in tetrads (Fig. 11, B and C, arrows), and tetrads are part of an array. The forming arrays of tetrads in transfected 1B5 cells are always located on a slightly raised membrane mound, suggesting that they are located at sites where the SR is closely opposed to the surface membrane, forming a peripheral coupling. Tetrads, in the domains of both normal skeletal muscle and transfected 1B5 cells, are often incomplete, that is they miss one or more components (see also Takekura et al., 1994; Franzini-Armstrong and Kish, 1995; Protasi et al., 1997 for similar observations in a variety of muscle fibers and in the BC₃H1 cell line). Examples of complete tetrads and tetrads with one missing element from transfected 1B5 cells are shown at the bottom right of Fig. 11. Complete tetrads, formed by four large and tall particles disposed at the corners of a square, are shown in Fig. 11, bottom right, top row. A minor distortion of the precise square arrangement is sometimes introduced by the fracturing process. Incomplete tetrads with three components are shown in the bottom row. These are unequivocally identified as tetrads with one missing particle, because the three particles are of the same size, because they subtend a 90° angle, and because they are part of a “domain of large particles.” Tetrads may be incomplete for different reasons: (a) one or more particle may break during fracturing and appear as a small stump (examples at left in the bottom row), or (b) components may be missing from the array, as shown in the last two examples at right in the bottom row (Protasi et al., 1997). Both situations are found in tetrads of normal myotubes developing in vivo and in vitro (see references above).

During normal differentiation of cardiac and skeletal muscle, RyRs, DHPRs, and triadin are always coclustered within calcium release units, giving the impression that these proteins and their interactions play a major role in the formation of the junctions between the SR and exterior membranes (Yuan et al., 1991; Flucher et al., 1993a,b, 1994; Protasi et al., 1996; for reviews see Franzini-Armstrong and Jorgensen, 1994; Flucher and Franzini-Armstrong, 1996). However, observations of various model systems reveal that docking of the SR to exterior membranes may occur in the absence of either DHPRs or RyRs (Franzini-Armstrong et al., 1991; Flucher et al., 1993b; Takekura et al., 1995b; Takeshima et al., 1995; Powell et al., 1996; Takeshima et al., 1997). Junctions are also formed in the muscle fibers of Ry₁R/Ry₃R double knockout mice (Ikemoto et al., 1997), showing that the presence of a small amount of Ry₃R is not needed for SR/ exterior membrane docking. In addition, in normal developing cardiac muscle, docking of the SR to the surface membrane precedes the localization of DHPRs and RyRs at the junctions (Protasi et al., 1996). Clustering of SR junctional proteins at calcium release units is also independent of the presence of DHPRs, since RyRs, triadin, and calsequestrin are present at SR–surface membrane junctions in dysgenic myotubes lacking α₁-DHPR (Franzini-Armstrong et al., 1991; Flucher et al., 1993b).

The findings in this study of dyspedic junctions add three further clues to the formation of calcium release units. The first is that DHPRs cluster at the junctions in the absence of RyRs. It must be assumed that other proteins of the junctional SR, perhaps even the same components that are responsible for SR–surface docking, have the role of holding DHPRs at the junctional sites. This is confirmed by the fact that CHO cells, which are of epithelial origin, seem to lack the component(s) needed for both docking of ER to the surface membranes and for extensive DHPR clustering, even when expression of skeletal RyR and a cardiac–skeletal DHPR chimera is induced in these cells (Takekura et al., 1995a).

A second result is that the presence of RyRs is not needed for the clustering of triadin in calcium release units. Two quite distinct roles have been proposed for triadin. On one hand, triadin is thought to reach out into the junctional gap between SR and surface membranes and either facilitate or directly participate in the joining of DHPRs to RyRs in skeletal muscle (Caswell et al., 1991; Fan et al., 1995a,b). On the other hand, the same molecule is described as being mostly located intraluminally, where it may help to link calsequestrin to the junctional SR membrane (Knudson et al., 1993a,b; Guo and Campbell, 1995). Our finding that triadin is present at the dyspedic junctions in the absence of RyRs does not help to solve the question of triadin function.

A third, and most important finding is that DHPRs do not associate into tetrads in the absence of Ry₁R (Figs. 11,A, and 12,A) and that tetrads are restored after “rescue” of the dyspedic phenotype by transfection of dyspedic cells with the Ry₁R cDNA (Figs. 11,C, and 12,B). These results provide direct confirmation of two current hypotheses on junctional architecture. One is that DHPRs are linked to RyRs in skeletal muscle and this linkage is responsible for the formation of tetrads (Block et al., 1988; Takekura et al., 1994; Protasi et al., 1997), and the other is that the random disposition of cardiac muscle DHPRs within junctional domains is, on the contrary, because of a lack of a link to the cardiac RyRs (Sun et al., 1995; see Fig. 12 C).

Present and past observations lead to the conclusion that the targeting of RyRs and DHPRs, as well as triadin and calsequestrin, to the junctional areas may be independent of each other and probably depend on other junctional components of a still unknown identity. However, one finding during the development of junctions in avian cardiac muscle seems at odds with this conclusion. Differentiation of the calcium release units in this muscle involves an initial docking of the SR to the surface, and then a coordinated assembly of groups of feet and DHPRs, which do not fill the entire docked area, but cluster together on one side (Protasi et al., 1996).

A previous study of dyspedic skeletal muscle in vivo agrees with the results presented here in showing formation of junctions that lack feet, but differs because clusters of large particles (DHPRs) were not detected in the plasma membrane of dyspedic muscle in mouse (Takekura et al., 1995b). The reason for this difference is not clear, since mRNA for the DHPR is present in the in vivo dyspedic muscle (Takeshima et al., 1994), and the protein is expressed (although in reduced amounts; ∼50% [Buck et al., 1997]). Also, charge movement is detected in primary cultures of dyspedic muscles (Nakai et al., 1996).

Muscle fibers of normal mouse express low but detectable levels of Ry₃R (Giannini et al., 1995). A few groupings of feet, presumably representing Ry₃R, have been detected in dyspedic myotubes developing in vivo (Takekura et al., 1995b), confirming functional evidence for caffeine-sensitive calcium release in primary dyspedic myotubes (Takeshima et al., 1995). In 1B5 cells, on the other hand, Ry₃R cannot be detected by biochemical approaches (Moore et al., 1998), and we have not found evidence for feet in thin sections from these cells suggesting that Ry₃R is not expressed in these cells at all.

1B5 and BC₃H1, two cell lines that express skeletal muscle–specific proteins of the junctional SR have three interesting structural defects in common: the cells either lack Z lines or have poorly developed ones; the nonjunctional sarcoplasmic reticulum is very poorly represented; peripheral couplings and junctions between SR and primitive T tubules on the other hand are relatively frequent. Since the nonjunctional SR develops early and is associated with the Z lines of myofibrils during normal skeletal muscle differentiation (Flucher et al., 1991, 1992, 1993a), the first two defects listed above may be causally linked.

The findings of this work highlight three important structural and developmental details of Ca²⁺ release units of muscle cells: the requirement of Ry₁R for the formation of tetrads, a structural feature of Ca²⁺ release units that is unique to skeletal muscle; the independence from RyRs of the association of triadin and DHPRs with the forming junctions; the independence from RyR of the initial docking of SR to exterior membranes. The restoration of tetrads in transfected 1B5 cells gives the ultimate demonstration that the DHPRs arrangement in tetrads is strictly dependent on their interaction, direct or not, with Ry₁R subunits.